Strongly disagree. Essentially you're entire logic is based around UConn failing. Essentially you're making decisions assuming the team can't convert and can't get a stop. The right call there is absolutely forcing Fresno to require both the TD and a successful conversion to take the lead. 20-14 does that.... neither 19-14 nor 18-14 does that.
Your logic is predicated not on doing what is best for UConn's odds of winning on that play (and makes it harder for Fresno to take the lead).. but instead, what would make it easier for UConn to take the lead only after losing the lead again. Essentially you're playing the game assuming your defense, which has been good can't stop them and it's going to take another possession to "really" win. The play there is to get the two, force Fresno into a situation where a TD only ties the game and they have to convert to take the lead.
For whatever it's worth, here's 538's analytic based "decision chart":
When To Go For 2, For Real
And here's the standard one that most coaches use:
Two Point Conversion Chart
In both sets of charts (analytic or standard), with a 4 point lead, the decision is to go for two.